Context & My role

Akka is a B2B fintech app connecting pre-ipo startups and potential investors. The users are investors that are searching for pre-ipo startups to invest in. 

I was working closely with the PMs, the dev and analytics teams.

My role was taking full ownership of the end-to-end referral journey, I had a strict 2-week design sprint. Relying on my experience with large-scale B2C products, I knew we needed a frictionless growth loop that focuses on consumer psychology. To meet the deadline without compromising quality, I worked heavily on component architecture and AI-accelerated rapid prototyping, ensuring a seamless handoff to the dev and analytics teams.

Problem Details

User growth was slower than planned, so the business decided to introduce a referral feature to accelerate it.

The core challenge I had to solve in the referral feature was balancing a strong incentive with absolute transparency. The business proposed a variable reward system, meaning users earned a randomized amount for each successful referral (look at business requirements below for the explanation). If we over-promised the maximum payout, users would feel cheated when they earned a lower tier.

Additionally, our users are serious investors. When they share a link with a peer, their personal reputation is on the line. The offer and the user flow had to be extremely clear and trustworthy. If it looked like a scam or was confusing, they simply wouldn’t risk sharing it.

Business Requirements:

• Honest Communication: Explain the variable reward (€50–€1,000 range) simply and honestly in the UI (knowing internally that the system is calibrated to a €100 average payout).

• Tracking: Ensure key data points and funnel statistics are trackable.

• Re-engagement: Design a way to catch and re-engage users who dropped off before finishing the referral.

Research & discovery

I started by analyzing competitor referral flows to see what was working and where they were failing. Based on those best practices, I used AI to build rapid prototypes of our flow and immediately tested them with our current users.

Key Insights from User Testing:

1. Uncertainty kills trust: Because our reward was variable, users hated vague promises. If they didn’t understand how the payout worked, they assumed it was a scam.

2. No time for follow-ups: Our users are busy investors. They are not going to manually track whether their friends signed up, and they do not have the time to explain the rules to them. The sharing process had to be effortless.

3. Complex steps cause drop-offs: When looking at competitors, the biggest failure was a confusing user journey. Users often couldn’t figure out the exact steps required to actually get the money, so they just abandoned the flow.

Explorations & the bet I made

Because the reward was a range and not a guaranteed exact amount, it created a new design challenge. I had to strike a very delicate balance. I wanted the “surprise” and “excitement” of potentially winning €1,000, but the UI had to remain sober. I could not let it evoke the aesthetics of a casino or a gambling app, because our users are serious investors.

The Bet I Made

My research showed that users hate uncertain money promises. My “bet” to the team was that we could still use this exciting variable reward structure if we presented it with extreme professional transparency.

To make this work, the goal of my final designs was to make the experience attractive, memorable, and motivating, while keeping the visuals completely anchored in the professional investment world.

Final solution

Offer screen: Shows the referral flow for a new user: nothing sent, earned, or pending.

Tracking the referrals: The user has shared and people start signing up: €100 earned, 3 pending.

Referee invite screen: The referee sees after tapping the shared link. It’s personalized (“Alex invited you”) to carry the trust of the referral.

design process

A strict 2-week deadline meant I had to work smart. To move fast without losing quality, I focused on three areas:

• Design System: I relied heavily on our existing component architecture. This kept the UI consistent and accessible. Reusing established patterns saved hours of design time.

• AI-Accelerated Workflow: I used AI to skip the tedious work. For example, I used Claude to write realistic copy for the share messages. Later in the sprint, when I realized I needed to map out all possible edge cases and error states quickly, AI allowed me to generate those layouts without missing a beat. Inside Figma, I used AI tools to quickly generate different layout options for the reward screens. This let me test multiple ideas fast and start user testing days early.

• Dev Handoff & Partnership: Design and engineering worked side-by-side. I structured my Figma files with a strict token system. These tokens perfectly matched the developers’ React code and utility classes. I also used AI plugins to check my files for missing variables before the final handoff. Because of this, developers did not have to guess my intentions. They could inspect the file, grab the exact specs, and build immediately.

Impact

What exactly drove the success of this feature? Design rationality. By choosing extreme transparency over flashy gamification, and by removing all friction for our busy users, the referral loop drove real business value:

• 600+ new verified investors acquired (~20% increase in our active community).

• 35% reduction in Customer Acquisition Cost (CAC) compared to paid ads.

• 28% conversion rate from referral-link click to paid subscription.

Next Iterations

We are already planning the next steps to optimize the funnel:

• Exit Survey: If a user abandons the offer screen before sharing, we will trigger a short prompt. It asks one simple question with three quick-tap reasons (e.g., “Not interested,” “Reward unclear,” “Will do it later”). This closes the feedback loop so we understand why users drop off, rather than just seeing where it happens.

• Smart Reminders: If a user starts the flow but does not share a link, we will send a targeted push notification reminding them of the reward they left on the table.
  
  

Reflection

With a strict 2-week deadline, it is very easy to fall into the “happy path” trap. I spent the first week perfecting the ideal scenario where a user successfully invites a friend. However, I underestimated how much the “unhappy paths” would impact user trust.

What happens if the friend’s invite link expires? What if the friend is already registered? I realized late in the sprint that if these error states were not designed perfectly, it would look like we were intentionally tricking the user out of their reward. I had to pivot quickly, using AI tools to help me rapidly map and design all the edge cases. It was a strong reminder that in fintech, trust is built in the error states, not just the happy paths.